Journal of Optoelectronics and Advanced Materials Vol. 4, No. 2, June 2002, p.

207 - 216

BULK AMORPHOUS MAGNETIC MATERIALS

The relationship between microstructure and magnetic properties of melt-spun ribbons with thicknesses up to 150 m and cast rods having up to 3 mm in diameter prepared by mould casting and suction casting, with nominal compositions Fe56Co7Ni7Zr6M1.5Nb2.5B20 (M = Zr, Ti, Ta or Mo) and Nd90-xFexSi10-yAly (x = 35 50; y = 0 10) was investigated. Saturation magnetizations up to 1.1 T, coercive fields of about 5 A/m, magnetic permeability of 2500030000 in the as-cast state were measured for Fe-based amorphous alloys. The large values of the intrinsic coercive field over 300 kA/m at room temperature and over 600 kA/m at 200 K, measured in a maximum field of 1 T for Nd-Fe-based "X-ray amorphous" alloys, and the strong dependence of the coercive field on temperature and cooling rate are ascribed to the development in the amorphous matrix of a metastable or non-equilibrium magnetic phase consisting in very small clusters whose sizes are below 5-10 nm. The magnetic properties of Nd-Fe-based bulk amorphous alloys are very sensitive to the Nd:Fe ratio, the addition element, the method of preparation, the thickness of the samples and the type of thermal or thermomagnetic treatment.

(Received February 5, 2002; accepted May 15, 2002)

1. Introduction The interest in glass-forming alloys, which vitrify at relatively low cooling rates (100 102 K/sec) from the molten state compared with conventional rapidly quenched metallic glasses, has grown in the last years. Owing to their resistance to crystallization, these easy glass-forming alloys can be cast in bulk shapes with dimensions of millimeters. The development of these new amorphous alloys may contribute widely to the understanding of the physical processes that take place in the supercooled liquid region and to the amorphization process itself due to the high thermal stability of the supercooled liquid region, making also possible the design of new applications based on their ability to be prepared in three-dimensional shapes. The history of bulk metallic glasses is dating back in time [1]. Despite of that, no ferromagnetic bulk metallic glasses were successfully synthesized before 1995. Inoue and co-workers from Tohoku University, Sendai, Japan announced in 1995 the preparation of the first bulk amorphous alloys with soft magnetic properties at room temperature in Fe-(Al,Ga)-(P,C,B,Si) multicomponent system [2,3] and one year later the first bulk amorphous hard magnetic alloys synthesized in (Nd,Pr)-Fe-Al system [4,5]. These alloys are very attractive for the subsequent development of the basic science and engineering applications, but so far as we know the mechanisms governing their magnetic behavior are not well known and understood. This paper is restricted to the magnetic properties of two families of bulk amorphous alloys that we consider representative for this field: soft magnetic (Fe,Co,Ni)-Zr-B based alloys and room temperature high-coercivity Nd-Fe-(Al,Si) alloys. Emphasis is placed on the interplay between the atomic structure and the magnetic properties. The influence of the composition, preparation technique (cooling rate), annealing and the on-set of crystallization on magnetic properties of Fe-based and NdFe-based bulk amorphous alloys is investigated. Brief remarks conclude the paper. Finally, some potential applications are suggested.

208

N. Lupu, H. Chiriac

2. Experimental Master alloys of 5-10 g each were prepared by arc melting pure elements (99.99 %) in an argon atmosphere. Ribbons with thicknesses between 20 and 200 m and widths of 2-5 mm were obtained by maintaining constant the crucible orifice size and the ejection pressure and changing the wheel velocity between 30 and 2.5 m/sec. Fe56Co7Ni7Zr6M1.5Nb2.5B20 (M = Zr, Ti, Ta, Mo) amorphous cast rods with diameters between 0.5 and 3 mm, Nd90-xFexAl10 (x = 20 50) amorphous cast rods with diameters of 1 and 1.5 mm and Fe56Co7Ni7Zr7.5Nb2.5B20 amorphous cast rings with Dint = 6 mm, Dext = 10 mm and t = 1 mm were prepared by Cu mould casting method. The method consists in applying an overpressure at the upper part of the molten alloy and forcing the melt to flow in a water-cooled Cu mould having different cavities within. Amorphous cylinders with diameters between 0.6 and 3 mm from Nd90-xFexAl10-ySiy alloys (x = 20 50; y = 0 10) were produced by suction casting method. The master alloy is arc melted in a water-cooled Cu crucible and sucked up thereafter in a cooled Cu mould having inside different shaped cavities. Qualitative information about the structure of Fe-based and Nd-Fe based ribbons and cast samples were obtained by X-ray diffraction (XRD) using Cu-K and Mo-K radiations. The structure was checked on both sides as well as on powders obtained by milling the ribbons, while XRD investigations on cast samples (rods or rings) were performed using powders obtained after crushing the rod into small pieces and then milling. Neutron diffraction measurements have been made on amorphous Nd90-xFexAl10 (x = 20 60) at 12 temperatures between 15 and 773 K. Total structure factors, S(Q) have been measured for melt-spun ribbons with thicknesses between 20 and 120 m and cast rods with diameters no larger than 2 mm by powder diffraction using the SLAD diffractometer at the Studsvik Neutron Research Laboratory. The data have been modeled using the reverse Monte Carlo (RMC) method [6] under the assumption that the atomic structure is completely disordered. In the reverse Monte Carlo simulations reported here, systems containing 2000 2500 particles were used. The input experimental data was S(Q). The experimental number density, , was 0.055 -3. Closest-approach distances were determined on the basis of the corresponding experimental data. All calculations were started from hard sphere random configurations. For reaching equilibrium (the closest fit to the experiment), 6.5-8.5 x 106 accepted moves had to be completed. Data about glass transition temperature (Tg), crystallization and melting temperature, thermal stability of the supercooled liquid region and the evolution of the crystallization process with the composition and samples thickness were acquired from calorimetric measurements (DSC) performed at a heating rate of 20 K/sec. The difference between Fe-based and Nd-Fe-based amorphous alloys consists in the position of the glass-transition temperature relative to the crystallization temperature: while for Fe-based, T g is situated below Tx, like in most of the previous studied amorphous alloys, for Nd-Fe-based Tg can not be evidenced on DSC curves, because it is located between Tx and T m and the large glass-forming ability is given by the reduced crystallization temperature (Tx/Tm). DC-magnetic measurements above room temperature but not exceeding 1100 K, at applied magnetic fields limited to 1260 kA/m, were carried out using a homemade vibrating sample magnetometer (VSM). The variation of the magnetization and coercive field below room temperature (5 300 K) was studied in fields up to 800 kA/m using a superconducting quantum interference device (SQUID) magnetometer and for maximum applied fields of 7.2 MA/m using an Oxford MagLab VSM. The magnetic field was applied in the axial direction of the samples. Each sample was thermally demagnetized prior to recording the magnetic measurements. The magnetic permeability (e) and the electrical resistivity (RT) of the Fe-based amorphous alloys were determined by using an AC fluxmetric method in a maximum applied field of 1000 A/m, while the saturation magnetostriction constant (s) was measured by the small angle magnetisation rotation (SAMR) method. 3. Results and discussions 3.1 Soft magnetic bulk amorphous alloys DSC curves exhibit one endothermic peak corresponding to the glass transition for Fe56Co7Ni7Zr6Nb2.5M1.5 B20 amorphous alloys. The existence of one exothermic peak for the

Bulk amorphous magnetic materials

209

amorphous alloys containing 1.5 at. % Zr, Ta or Mo and a shifted peak for Fe56Co7Ni7Zr6Nb2.5Ti1.5B20 amorphous alloy indicates one stage crystallization process for the former alloys and two-stage crystallization for the last one [7]. The substitution of 1.5 at. % Zr with Ti, Ta or Mo leads to the increase of the resistance of the supercooled liquid region against crystallization and consequently the maximum thickness for which the bulk samples are still amorphous increases about 2 times, from 1.6 mm for M = Zr to 3 mm for M = Ti, Ta and Mo. Although the replacement of Zr with Ti, Ta or Mo causes the decrease of the glass transition temperature, Tx increases with 10 degrees from 80 to 90 K [8]. Another evidence for the formation of the amorphous phase in (Fe,Co,Ni)70(Zr,M,Nb)10B20 alloys is given by X-ray diffraction patterns, which indicate just one broad maximum characteristic to the amorphous phase and no sharp lines corresponding to crystalline phases for thick ribbons 150 m in thickness, regardless of their composition. The fully amorphous structure was also found in Fe56Co7Ni7Zr7.5Nb2.5B20 bars with diameters up to 1.6 mm and rings with t = 1 mm as well as in bars having diameters up to 3 mm prepared from Fe56 Co7Ni7Zr6Nb2.5M1.5B20 (M = Ti,Ta,Mo) alloys. The existence of a single amorphous phase and no crystalline traces in the Fe56Co7Ni7Zr6 M1.5Nb2.5B20 bulk amorphous samples is evidenced also by thermomagnetic measurements [9]. The first decrease of the magnetization to zero with the temperature increase gives the value of the Curie temperature of the amorphous phase (about 554 K) and proves the amorphicity of the sample. The magnetization starts to increase by the on-set of crystallization (about 860 K), reaches a maximum and decreases again to zero when the fully crystalline sample becomes paramagnetic, indicating a value around 1100 K for the Curie temperature of the crystalline alloy. The magnetization increases when the sample suffers a phase transition from amorphous to crystalline. The change of the composition of the Fe56 Co7Ni7Zr7.5Nb2.5B20 basic alloy causes not just the increase of the glass-forming ability shown previously, but also the modification of the soft magnetic properties as it can be seen in Table 1. While the saturation magnetization does not change significantly by introduction of Ti, Ta or Mo instead of Zr in the basic alloy, the magnetic permeability decreases from 21500 for the Fe56 Co7Ni7Zr7.5Nb2.5B20 amorphous ribbons having 35 m in thickness to 17000 for the Fe56Co7Ni7Zr6Nb2.5Mo1.5B20 amorphous ribbons with the same thickness. The Curie temperature increases and coercive field decreases by the substitution of Zr with Ti, Ta or Mo.
Table 1. Soft magnetic characteristics of the Fe56Co7Ni7Zr6Nb2.5M1.5B20 (M=Zr,Ti,Ta,Mo) amorphous ribbons having 35 m in thickness.

0Ms

M = Zr M = Ta M = Ti M = Mo

(T) 1.01 0.89 1.06 1.07

Hc (A/m) 9.5 6.06 6.1 7.43

This behavior could also be explained on the basis of the increase of the thermal stability of the supercooled liquid region. The crystallization energy, determined by the area under the exothermic peak on DSC curves, is larger for amorphous alloys with Ti and Mo suggesting a higher degree of structural disorder and consequently the development of different magnetic structures more stable. The large values of the magnetic permeability obtained for Fe56Co7Ni7 Zr6M1.5 Nb2.5B20 amorphous alloys are closely related to the high electrical resistivities (RT) and to the relatively small saturation magnetostriction constant (s) measured in the as-cast state. RT increases from 1.5110-8 m for M = Zr to 2.0810-8 m; 1.8710-8 m and 1.7610-8 m for M = Ti, Ta and Mo, respectively. s ranges from 710-6 for Fe56Co7Ni7Zr7.5Nb2.5B20 amorphous alloys to 1010-6 for the amorphous alloys having Ti, Ta or Mo in composition. The small values of s explain also the reduced values of the coercive field. The improvement of the magnetic properties of Fe56Co7Ni7Zr6M1.5Nb2.5B20 amorphous alloys is obtained by choosing the suitable composition or by increasing the thickness of the samples, namely by attaining a high degree of structural relaxation in the as-cast state. Fig. 1 shows the behavior of the magnetic permeabilities at 500 Hz as a function of thickness for

(500 Hz) 21,500 18,000 19,000 17,000

Tc (K) 554 560 603 560

210

N. Lupu, H. Chiriac

Fe56Co7Ni7Zr7.5Nb2.5B20 melt-spun ribbons and cast rods. The increase of the amorphous ribbons thickness leads to the structural relaxation and consequently the permeability increases. The thicker amorphous ribbons the higher degree of structural relaxation and the better magnetic properties. The role of the eddy current in the amorphous ribbons with thicknesses up to 150 m is very small and hence it could be neglected. The on-set of crystallization in ribbons with thicknesses larger than 150 m results in the decrease of the magnetic permeability and in the displacement of the maximum of the permeability towards high-applied fields. The magnetic permeability of the Fe56Co7Ni7Zr7.5Nb2.5B20 amorphous cast rods with diameters up to 1.6 mm is smaller than the values obtained for melt-spun amorphous ribbons. This behavior is related to the development of different magnetic domain structures in rods in comparison with ribbons due to the different cooling rates used in the preparation process and the difference between the amorphous samples dimensions. For Fe56Co7Ni7Zr7.5Nb2.5B20 cast rod 3 mm in diameter, which is crystalline in the as-cast state, the magnetic permeability is very small. Thus, the good soft magnetic properties of Fe56Co7Ni7Zr6M1.5Nb2.5 B20 bulk amorphous alloys are strongly related to the existence of the amorphous phase.
3.0x10 2.5x10 2.0x10 1.5x10 1.0x10 5.0x10
4

rod d = 0.5 mm rod d = 1.6 mm rod d = 3 mm

ribbon 25 m x 1 mm ribbon 60 m x 1 mm ribbon 165 m x 1 mm

0.0 1 10 Applied Field (A/m) 100

Fig. 1. Magnetic permeabilities (at 500 Hz) of Fe56Co7Ni7Zr7.5Nb2.5B20 melt-spun ribbons and cast rods.

It is important to notice that at 10 kHz the magnetic permeability is the same for Fe56Co7Ni7Zr6M1.5Nb2.5 B20 amorphous alloys in the shape of ribbons and bulk samples (rods or rings), while at 500 Hz they differ with about 35 %. This observation suggests bulk amorphous alloys as good for application at intermediary frequencies. Structural relaxation of Fe56Co7Ni7Zr6M1.5Nb2.5B20 amorphous alloys can be also reached by annealing the amorphous samples at temperatures below glass transition temperature (Tg). Treatments applied below 753 K (20 degrees below Tg) leave the sample in the amorphous state and lead to the increase of the magnetic permeability while the magnetization remains almost the same as in the ascast state. The subsequently increase of the annealing temperature results in the on-set of the nucleation leading to the decrease of the magnetic permeability and the increase of the magnetization. In the same time by increasing the annealing temperature the magnetization decreases but doesn't become zero, indicating a mixture of amorphous and crystalline phases. The coercive field exhibits initially a decrease with temperature and starts to increase when annealing temperature exceeds Tx, this increase being much pronounced for the basic alloy (M = Zr). The precipitation of crystalline phases results in suddenly increase of the coercive field (Hc). Fig. 2 shows the modification of the shape of the hysteresis loops after thermal and thermomagnetic treatments for one thick amorphous ribbon (120 m) prepared from the basic alloy Fe56Co7Ni7Zr7.5Nb2.5B20. The direction of the applied magnetic field is relative to the long axis of the ribbon.

Bulk amorphous magnetic materials

211

0.50
as-cast 300" / 748 K without Hext 300" / 748 K with Hext // 300" / 748 K with Hext

0.25

0Ms (T)

0.00

-0.25

-0.50 -20

-10

0 Applied Field (A/m)

10

20

Fig. 2. The hysteresis loop shape dependence on the type of treatment for Fe56Co7Ni7Zr 7.5Nb2.5B20 amorphous alloys.

One observes that the rectangularity of the hysteresis loop as well as the saturation magnetization and coercive field are changing with the direction of the applied magnetic field. While the thermal treatment without magnetic field below Tg (= 773 K) leads to the slight increase of the coercive field and to the elongation of the hysteresis loop, the presence of an external magnetic field results in better magnetic properties, i.e. the better softness of the amorphous material. The best response is obtained when the applied field is parallel with the long axis of the ribbon because of the easy axis induced on the parallel direction with the applied field. When the external field is perpendicular to the ribbon long axis the induced anisotropy is smaller owing to the competition of the magnetoelastic anisotropy induced during the melt-spinning preparation process and oriented parallel with the long axis. Consequently, the coercive field increases. Further experimental work along this line is expected in the near future.

3.2 High-coercivity bulk amorphous alloys The large glass-forming ability of Nd90-xFex(Al,Si)10 alloys is due to the high values of 0.8 to 0.92 of the reduced crystallization temperature (T x/Tm, where Tx and Tm are crystallization and eutectic melting temperature, respectively). DSC curves indicate only one exothermic peak due to crystallization and neither endothermic reaction due to the glass transition nor the supercooled liquid region in the temperature range before crystallization [9]. The total replacement of Al with Si results in the decrease of the glass-forming ability and consequently the ability to fabricate bulk cast samples with fully amorphous structure is decreased. The increase of the Nd content (for 10 at. % Al), results in the enlargement of the glass-forming ability. The crystallization energy (the area under the exothermic peak) is higher for the melt-spun ribbons containing a higher amount of Fe in comparison with the rods due to the more disordered amorphous structure developed in the first ones. The crystallization energy doesnt change significantly as a function of the Fe content for Nd90-xFexAl10 thin ribbons 30 m thickness, whereas for cast rods it strongly decreases with the increase of the Fe content as a consequence of the decrease of the glass-forming ability. This behavior is explained by the insufficient cooling rates assured by the casting techniques for the alloys with larger contents of Fe and indicates a strong correlation between the composition, the cooling rate and the glass-forming ability for Nd-Fe-based ternary amorphous alloys. In Fig. 3 are presented the neutron diffraction patterns taken at room temperature for Nd90-xFexAl10 (x = 35; 40; 50) melt-spun ribbons with thicknesses of 25 and 120 m, respectively. Due to the decrease of the glass-forming ability, the Nd40Fe50Al10 ribbon 120 m, cannot be obtained fully amorphous.

212

N. Lupu, H. Chiriac

3.5 3.0 2.5 S (Q) 2.0 1.5 1.0 0.5 Nd55Fe35Al10

1.2
25 m 120 m

0.8

0.4 G (r)

0.0

-0.4

-0.8

6 Q ( )
-1

10
2.0

10

11

r ()

3.5 3.0 2.5 S (Q) 2.0 1.5 1.0 0.5 Nd50Fe40Al10 25 m 120 m
G(r)

1.5 1.0 0.5 0.0 -0.5 -1.0

25 m 120 m

6 Q ( )
-1

10

10

11

r ()

5.5 5.0 4.5 4.0 3.5 S(Q) 3.0 2.5 2.0 1.5 1.0 0.5 0.0 0 2 4 6 Q ( )
-1

1.0

Nd40Fe50Al10
25 m 120 m
G (r)

0.8 0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6 -0.8

25 m

10

10

11

r ()

Fig. 3. Neutron diffraction pattern as a function of wavevector , Q = 4 sin reduced radial distribution function G ( r ) = 4r
2

and

(r ) for Nd90-xFexAl10 melt-spun ribbons.

The split first peak and the pronounced structure at higher angles are features not present in the patterns for the other amorphous alloys with dense random packing disordered structures but they were observed in Nd-Fe binary amorphous alloys [10]. The split is more pronounced for the thicker ribbon than for the thin ones. We suppose this behavior being related to the more relaxed structure developed in thick ribbons. The reduced radial distribution functions, G(r), obtained by Fourier transform of the total structure factor S(Q), show first neighbors peaks at 2.54, 2.85, 3.36 and 4.20 . The peak at 2.54 is at the position expected for the nearest neighbor Fe atoms in the dense random packing (DRP) model; the other peaks cannot be correlated to combinations of the radii of Nd (1.82 ), Al (1.43 ) or Fe (1.27 ) atoms as predicted by the DRP model. This disagreement, which is

Bulk amorphous magnetic materials

213

particular to the case of the present light rare-earth RETM alloys, could be ascribed to the development of a new type of disordered structure. Thus, one suggests that the structure consists in a dense random packing of nanometer-sized atomic clusters, the size being dependent on the thickness of the ribbon, i.e. the melt-spinning conditions. Lets consider atoms like rigid spheres dense randomly packed in a given structure. If we ignore the intensity of the chemical bonding, thus the interatomic distances are equal with the sum of the atomic radii. Considering rNd = 1.82 , rFe = 1.27 and rAl = 1.43 , the theoretical values for Nd-Nd, Nd-Fe, Nd-Al, Fe-Fe, Fe-Al and Al-Al atomic pairs are respectively: 3.64 , 3.09 , 3.25 , 2.54 , 2.7 and 2.86 . Comparing these values with those calculated from experimental data (Table 2) one observes that Fe-Fe values fit very well, whereas the others are larger than experimental ones. This can be explained in the hypothesis of the existence of randomly distributed Fe clusters, while Nd and Al atoms are distributed randomly between these clusters.
Table 2. The interatomic distances (r) and the coordination numbers (N) as a function of x, for Nd50Fe40Al10 melt-spun ribbons, at room temperature.

Fe-Fe Thickness, t (m) 25 120 r () 2.54 2.51 N 8.65 7.69

Fe-Nd(Al) r () 2.85 2.76 N 4.85 7.39

Nd-Nd r () 3.44 3.38 N 5.72 8.30

Consequently, one expects complex magnetic behavior dependent on thickness and composition. Values of the magnetization measured in a field of 800 kA/m and coercive field obtained at room temperature for Nd50Fe40 Al10 amorphous samples are given in Table 3. The large values of the coercive field can be explained by assuming the existence of a microstructure consisting of very small magnetic clusters, whose size approaches a single magnetic domain, dispersed in the amorphous matrix, in agreement with the neutron diffraction results as well as with the previous results obtained for melt spun Nd-Fe amorphous alloys [11]. The size and the composition of these magnetic clusters are strongly dependent on the cooling rate. The sizes of these magnetic clusters are small enough and cannot be evidenced by XRD measurements.
Table 3. Magnetic characteristics of Nd50Fe40Al10 amorphous ribbons and rods, at room temperature.

Sample Amorphous ribbon Amorphous ribbon Amorphous ribbon Amorphous ribbon Amorphous ribbon Amorphous rod Amorphous rod

Thickness (mm) 0.030 0.055 0.085 0.105 0.142 0.6 1.8

0M {10 kOe) (T) 0.33 0.24 0.25 0.32 0.23 0.25 0.24

iHc

(kA/m) 64 123 226 274 286 244 246

The higher values obtained for the coercive field of the thick amorphous ribbons comparatively with thin ones are related to the more relaxed microstructure developed in the first ones. It is worthwhile to note that the magnetization for the ribbon of 105 m is almost the same like for the ribbon 30 m, whereas the coercive field is 4 times larger. This behavior can be explained by the largest degree of structural disorder existent in the ribbon of 105 m, in agreement with the largest crystallization energy obtained from DSC curves. In these conditions, an external applied field causes the alignment of more magnetic domains relative to its direction. The microstructure is more homogenous in amorphous rods due to their dimensions and lower cooling rates and consequently the values obtained for coercive field are smaller than for the thick ribbons.

214

N. Lupu, H. Chiriac

Nd90-xFexAl10 (x = 35 50) amorphous alloys exhibit large coercive fields at room temperature, even for low applied fields, regardless of their thickness and preparation techniques. The magnetization increases about 2 times by increasing the Fe content from 35 to 50 at. %, regardless of the amorphous samples shape, the same behavior being observed for high fields up to 9 T. The coercive field presents a strong dependence on the composition. The on-set of crystallization in the Nd40Fe50Al10 thick ribbon results in the decrease of the magnetization, while the coercive field has almost the same value as that obtained for fully amorphous thick ribbons with a less content of Fe because of the existence of the amorphous residual phase. Thus, the large coercive fields of Nd-Febased ternary bulk amorphous alloys are related to the existence of the amorphous phase, its disappearance leading to the drastically decrease of the coercive field and moreover, to the disappearance of the ferromagnetic properties, as we will see in the following. The decrease of the Fe content results in the decrease of the coercive field of about 2 times for thin amorphous ribbons and in its increase of about 2 times for amorphous cast rods. For thick amorphous ribbons the coercive field is almost the same, regardless of the Fe content. For higher external fields (larger than 3 T) the coercive field doesnt change significantly with the Fe content for each different type of ribbon or for cast rods. The difference in the absolute value of the coercive field as a function of the amorphous samples' thickness results from the volume rate between Fe-based magnetic clusters and the homogenous Nd-rich matrix through the cooling rate. Thus, Nd90-xFexAl10-ySiy bulk amorphous alloys are structurally glassy but magnetically granular. The granular magnetic structure consists in Fe-based nanosized clusters dispersed in amorphous Nd-rich matrix. Due to their very small dimensions they cannot be evidenced by X-ray diffraction measurements, but their presence is proved by the specific magnetic behavior of these bulk amorphous alloys, especially by the high coercive fields in the as-quenched state and the dependence on the cooling rate. The composition and size of the magnetic clusters as well as the composition of the matrix in which the clusters are embedded are very sensitive to the preparation conditions, mainly the cooling rate and the thermal history of the molten alloy. Fe-based magnetic clusters are coupled between them through the amorphous matrix and the exchange coupling is strongly influenced by the magnitude of the Nd3+ single ion local anisotropy. Whereas at low temperatures the anisotropy plays the predominant role in the macroscopic magnetic response of the system, the increase of the temperature diminishes its effect and the main role is attributed to the ferromagnetic exchange energy. At low temperature, the local anisotropies of the Nd3+ ions are very strong and oriented randomly, thus they can freeze randomly the magnetic moments of Fe and Nd. The anisotropy axis corresponding to the Fe-Nd pairs are oriented randomly, and a small external field is not enough to align them. The AC-susceptibility measurements as well as the variation of the coercive field on temperature in the range 4 300 K [12] indicate magnetic transitions that suggest different local arrangements of the Nd atoms, similar to dhcp (double hexagonal compact) and fcc crystalline phases. The first transition appears around 10 K and seems to be related to the Nel transition temperature of the Nd antiferromagnetic. Another magnetic transition is observed around 70 K and is due to the transition from a completely disordered magnetic phase to a sperrimagnetic non-collinear one. Different magnetic structures can be obtained depending on temperature and the predominant coupling between Fe and Nd atoms. This is probably the explanation for the lack of the hysteresis loop at temperatures between 10 and 100 K and low external fields (< 1 T) [12]. In this range of temperatures Nd-rich matrix and Fe-based magnetic clusters are completely uncoupled, but the exchange coupling within the clusters still exist. If from any reason, the coupling between clusters is not possible, the macroscopic response of the system is weakly ferromagnetic. The maximum of the coercive field is attained at that temperature at which the anisotropy energy and the exchange energy have similar values, i.e. the magnetic clusters have the size of the single magnetic domain [13]. Taking in consideration the critical size for the Fe single domain of about 1000 and the strength of the local anisotropy of the Nd3+ ions results that the Fe magnetic moment is not aligned on a direction but forming a cone with an angle of about 300 [14]. Thus, the magnetic structures are non collinear and the critical size of the single magnetic domain is smaller than 100 nm. In Nd-Fe-based thin amorphous ribbons, due to their reduced dimensions, high cooling rate (which does not allow the structural relaxation) and thermal effects, the system tends to a superparamagnetic regime at room temperature, and consequently the coercive filed is not very large.

Bulk amorphous magnetic materials

215

It is known that in materials containing single magnetic domains the magnetization is due mainly by the magnetic moments rotations. If the value of the applied field is not large enough, the measured hysteresis loops are minor loops, and the measured magnetization is far from its saturation. This effect is more pronounced in materials in which the magnetic anisotropies play an important role in magnetization processes. Nd-Fe-based bulk amorphous materials with strong local random magnetic anisotropy of the Nd3+ ions, which becomes stronger by decreasing the temperature, belong to this class of materials. A proof of the role of the magnitude of the applied field on the magnetic response of the system is shown in Fig. 4 for Nd50Fe40 Al10 amorphous ribbons 25 and 120 m in thickness.
120
9T

Ribbon 25 m

140

9T 7T

5
Ribbon 120 m

5
120

100 80 M (Am /kg) 60 40 20 0 4

100 M (Am /kg)

5T

3
1T 3T 5T 7T 9T

Hc max (T)

5T

60 40 20 0
1T

3T

1T 5T 7T 9T

1T

1 300

50

100

150

200

250

50

100

150

200

250

0 300

Temperature (K)

Temperature (K)

Fig. 4. The magnetization and the maximum of the coercive field versus temperature and applied field for Nd50Fe40Al10 amorphous melt-spun ribbons (the coercive field is represented by symbols).

As larger the applied field as smaller the temperature at which the coercive field attains its maximum. One observes also that the magnetization exhibits different behavior for applied fields no large enough to align the magnetic moments parallel to its direction, i.e. to be large enough to overload the anisotropy energy given by the Nd3+ ions anisotropy. Thermomagnetic measurements (after cooling without field or after cooling in an external field) at temperatures above 300 K indicate the co-existence of two types of magnetic order: spinglass-like or cluster-glass-like short-range order within the clusters and long-range order ferromagnetic order like a response of the exchange coupling between clusters [15]. Consequently, the magnetic structure consists in randomly distributed non-collinear short-range magnetic ordered regions. 4. Perspectives Despite the considerable progress achieved in recent years concerning the knowledge of the atomic structure of amorphous alloys and concerning the understanding of their basic magnetic properties, several unsolved problems were encountered in the field of bulk magnetic amorphous alloys. There are still many questions related to the microstructure of these materials and its interplay with magnetic properties, especially for Nd-Fe-based high coercivity bulk amorphous alloys. The magnetic ground states of Nd-Fe-based cluster amorphous alloys and non-collinear structures existent in these materials are far from being fully characterized. These problems and others make the study of the magnetic properties of bulk amorphous alloys a fascinating field of research. Bulk magnetic amorphous alloys are interesting also for the field of engineering applications. References [1] H. S. Chen, Acta Metall., 22, 1505 (1974). [2] A. Inoue, J. S. Gook, Mater. Trans., JIM, 36, 1180, 1282 (1995). [3] A. Inoue, Y. Shinohara, J. S. Gook, Mater. Trans., JIM, 36, 1427 (1995).

Hc max (T)

80

216

N. Lupu, H. Chiriac

[4] A. Inoue, T. Zhang, W. Zhang, A. Takeuchi, Mater. Trans., JIM, 37, 99 (1996). [5] A. Inoue, T. Zhang, A. Takeuchi, W. Zhang, Mater. Trans., JIM, 37, 627 (1996). [6] D. A. Keen, R. L. McGreevy, R. I. Bewley, R. Cywinski, Nucl. Instrum. Meth. Phys. Res., A354, 48 (1995). [7] H. Chiriac, N. Lupu, J. Non-Cryst. Solids, 250-2, 751 (1999). [8] H. Chiriac, N. Lupu, J. Magn. Magn. Mater., 215-216, 394 (2000). [9] H. Chiriac, N. Lupu, Physica B: Condens. Matter, 299, 293 (2001). [10] H. A. Alperin, W. R. Gillmor. S. J. Pickart, J. J. Rhyne, J. Appl. Phys., 50, 1958 (1979). [11] J. J. Croat, IEEE Trans. Magn., MAG-18, 1442 (1982). [12] H. Chiriac, N. Lupu, K. V. Rao, R. E. Vandenberghe, IEEE Trans. Magn., 37 (2001) 2509. [13] B. D. Cullity, Introduction to Magnetic Materials, Addison_Wesle Publishing Company (1972) p. 385. [14] K. Moorjani, J. M. D. Coey, Magnetic Glasses, Elsevier, Amsterdam (1984). [15] H. Chiriac, N. Lupu, K. V. Rao, Scripta Mater., 44, 1379 (2001).